Tencent has developed WeKnora, an open-source, LLM-powered framework designed for enterprise document understanding, semantic retrieval, and reasoning. The system integrates three core functional modes: RAG-based Q&A, autonomous agents for multi-step tasks, and a structured "Wiki Mode" that utilizes agents to transform raw documents into interlinked Markdown pages via knowledge graphs. WeKnora supports over 10 document formats (including PDF, Word, and XMind) and features automated synchronization from diverse data sources such as Confluence, GitLab, Notion, and Feishu. Technically, the platform offers high modularity, allowing users to swap LLMs (supporting 27 vendors including OpenAI, DeepSeek, and Qwen), vector databases (such as Milvus, Qdrant, or pgvector), and storage backends. Key technical features include cross-session long-term memory, multi-workspace RBAC with four distinct roles, and a built-in MCP server for integration with tools like Cursor and Claude. For advanced agentic capabilities, it incorporates "BrowserSkill" to allow agents to operate local Chrome or Edge browsers and utilizes Docker, E2B, or Cube for sandboxed skill execution. Observability is managed through Langfuse, providing tracing for agent reasoning steps, token usage, and parsing pipelines. Deployment is flexible, supporting local Docker Compose, Kubernetes via Helm, or self-hosted environments.